We propose ordering-based approaches for learning the maximal ancestral graph (MAG) of a structural equation model (SEM) up to its Markov equivalence class (MEC) in the presence of unobserved variables. Existing ordering-based methods in the literature recover a graph through learning a causal order (c-order). We advocate for a novel order called removable order (r-order) as they are advantageous over c-orders for structure learning. This is because r-orders are the minimizers of an appropriately defined optimization problem that could be either solved exactly (using a reinforcement learning approach) or approximately (using a hill-climbing search). Moreover, the r-orders (unlike c-orders) are invariant among all the graphs in a MEC and include c-orders as a subset. Given that set of r-orders is often significantly larger than the set of c-orders, it is easier for the optimization problem to find an r-order instead of a c-order. We evaluate the performance and the scalability of our proposed approaches on both real-world and randomly generated networks.

Approximating integrals with respect to a complicated, highdimensional probability distribution π is an important problem spanning multiple disciplines, such as Bayesian statistical inference, machine learning, statistical physics, and theoretical computer science [13, 17]. Sequential Monte Carlo (SMC) methods are a large class of stochastic approximation algorithms designed to solve these problems by combining Markov chain Monte Carlo (MCMC) methods and resampling strategies to sequentially sample from a series of probability distributions. Some examples of SMC algorithms include population Monte Carlo methods [2], annealed importance sampling [27], sequential particle filters [3], and population annealing [40], among many others [15]. Closely related - but purely MCMC - methods include parallel tempering (PT) [14] and simulated tempering (ST) [23], which have been referred to as population-based MCMC methods [15]. An SMC sampler is generally constructed as follows.

The transferability of adversarial examples (AEs) across diverse models is of critical importance for black-box adversarial attacks, where attackers cannot access the information about black-box models. However, crafted AEs always present poor transferability. In this paper, by regarding the transferability of AEs as generalization ability of the model, we reveal that vanilla black-box attacks craft AEs via solving a maximum likelihood estimation (MLE) problem. For MLE, the results probably are model-specific local optimum when available data is small, i.e., limiting the transferability of AEs. By contrast, we re-formulate crafting transferable AEs as the maximizing a posteriori probability estimation problem, which is an effective approach to boost the generalization of results with limited available data. Because Bayes posterior inference is commonly intractable, a simple yet effective method called MaskBlock is developed to approximately estimate. Moreover, we show that the formulated framework is a generalization version for various attack methods. Extensive experiments illustrate MaskBlock can significantly improve the transferability of crafted adversarial examples by up to about 20%.

Attention-based deep learning models, such as Transformers (Vaswani et al., 2017; Devlin et al., 2019), have achieved unprecedented empirical success in a wide range of cognitive tasks, in particular in natural language processing (NLP). On the other hand, deep variational Bayesian approaches to representation learning, such as variational autoencoders (VAEs) (Kingma and Welling, 2014), have also been very influential, especially due to their variational information bottleneck (VIB) (Alemi et al., 2017; Kingma and Welling, 2014) for regularising the induced latent representations. Previous VIB methods only apply to a vector space, and Transformers crucially do not use a single vector as their latent representation, instead using a set of vectors (Lin et al., 2020; Fang et al., 2021; Park and Lee, 2021). This allows the number of vectors in a Transformer embedding to grow with the size of the input, which is essential for embedding natural language text (Bahdanau et al., 2015), where the size of the input can range from a single word to thousands of words. In this paper, we propose a variational information bottleneck regulariser for set-of-vector latent representations, and use it to regularise the induced latent representation of a Transformer encoder-decoder variational autoencoder.

We consider the application of the factor graph framework for symbol detection on linear inter-symbol interference channels. Based on the Ungerboeck observation model, a detection algorithm with appealing complexity properties can be derived. However, since the underlying factor graph contains cycles, the sum-product algorithm (SPA) yields a suboptimal algorithm. In this paper, we develop and evaluate efficient strategies to improve the performance of the factor graph-based symbol detection by means of neural enhancement. In particular, we consider neural belief propagation and generalizations of the factor nodes as an effective way to mitigate the effect of cycles within the factor graph. By applying a generic preprocessor to the channel output, we propose a simple technique to vary the underlying factor graph in every SPA iteration. Using this dynamic factor graph transition, we intend to preserve the extrinsic nature of the SPA messages which is otherwise impaired due to cycles. Simulation results show that the proposed methods can massively improve the detection performance, even approaching the maximum a posteriori performance for various transmission scenarios, while preserving a complexity which is linear in both the block length and the channel memory.

The problem of maximum likelihood decoding with a neural decoder for error-correcting code is considered. It is shown that the neural decoder can be improved with two novel loss terms on the node's activations. The first loss term imposes a sparse constraint on the node's activations. Whereas, the second loss term tried to mimic the node's activations from a teacher decoder which has better performance. The proposed method has the same run time complexity and model size as the neural Belief Propagation decoder, while improving the decoding performance by up to $1.1dB$ on BCH codes.

Spoken language recognition (SLR) refers to the automatic process used to determine the language present in a speech sample. SLR is an important task in its own right, for example, as a tool to analyze or categorize large amounts of multi-lingual data. Further, it is also an essential tool for selecting downstream applications in a work flow, for example, to chose appropriate speech recognition or machine translation models. SLR systems are usually composed of two stages, one where an embedding representing the audio sample is extracted and a second one which computes the final scores for each language. In this work, we approach the SLR task as a detection problem and implement the second stage as a probabilistic linear discriminant analysis (PLDA) model. We show that discriminative training of the PLDA parameters gives large gains with respect to the usual generative training. Further, we propose a novel hierarchical approach where two PLDA models are trained, one to generate scores for clusters of highly-related languages and a second one to generate scores conditional to each cluster. The final language detection scores are computed as a combination of these two sets of scores. The complete model is trained discriminatively to optimize a cross-entropy objective. We show that this hierarchical approach consistently outperforms the non-hierarchical one for detection of highly related languages, in many cases by large margins. We train our systems on a collection of datasets including over 100 languages, and test them both on matched and mismatched conditions, showing that the gains are robust to condition mismatch.

Exact inference in Bayesian networks is intractable and has an exponential dependence on the size of the largest clique in the corresponding clique tree (CT), necessitating approximations. Factor based methods to bound clique sizes are more accurate than structure based methods, but expensive since they involve inference of beliefs in a large number of candidate structure or region graphs. We propose an alternative approach for approximate inference based on an incremental build-infer-approximate (IBIA) paradigm, which converts the Bayesian network into a data structure containing a sequence of linked clique tree forests (SLCTF), with clique sizes bounded by a user-specified value. In the incremental build stage of this approach, CTFs are constructed incrementally by adding variables to the CTFs as long as clique sizes are within the specified bound. Once the clique size constraint is reached, the CTs in the CTF are calibrated in the infer stage of IBIA. The resulting clique beliefs are used in the approximate phase to get an approximate CTF with reduced clique sizes. The approximate CTF forms the starting point for the next CTF in the sequence. These steps are repeated until all variables are added to a CTF in the sequence. We prove that our algorithm for incremental construction of clique trees always generates a valid CT and our approximation technique preserves the joint beliefs of the variables within a clique. Based on this, we show that the SLCTF data structure can be used for efficient approximate inference of partition function and prior and posterior marginals. More than 500 benchmarks were used to test the method and the results show a significant reduction in error when compared to other approximate methods, with competitive runtimes.

We propose a novel model selection algorithm based on a penalized maximum likelihood estimator (PMLE) for functional hidden dynamic geostatistical models (f-HDGM). These models employ a classic mixed-effect regression structure with embedded spatiotemporal dynamics to model georeferenced data observed in a functional domain. Thus, the parameters of interest are functions across this domain. The algorithm simultaneously selects the relevant spline basis functions and regressors that are used to model the fixed-effects relationship between the response variable and the covariates. In this way, it automatically shrinks to zero irrelevant parts of the functional coefficients or the entire effect of irrelevant regressors. The algorithm is based on iterative optimisation and uses an adaptive least absolute shrinkage and selector operator (LASSO) penalty function, wherein the weights are obtained by the unpenalised f-HDGM maximum-likelihood estimators. The computational burden of maximisation is drastically reduced by a local quadratic approximation of the likelihood. Through a Monte Carlo simulation study, we analysed the performance of the algorithm under different scenarios, including strong correlations among the regressors. We showed that the penalised estimator outperformed the unpenalised estimator in all the cases we considered. We applied the algorithm to a real case study in which the recording of the hourly nitrogen dioxide concentrations in the Lombardy region in Italy was modelled as a functional process with several weather and land cover covariates.

Extracting governing physics from data is a key challenge in many areas of science and technology. The existing techniques for equations discovery are dependent on both input and state measurements; however, in practice, we only have access to the output measurements only. We here propose a novel framework for learning governing physics of dynamical system from output only measurements; this essentially transfers the physics discovery problem from the deterministic to the stochastic domain. The proposed approach models the input as a stochastic process and blends concepts of stochastic calculus, sparse learning algorithms, and Bayesian statistics. In particular, we combine sparsity promoting spike and slab prior, Bayes law, and Euler Maruyama scheme to identify the governing physics from data. The resulting model is highly efficient and works with sparse, noisy, and incomplete output measurements. The efficacy and robustness of the proposed approach is illustrated on several numerical examples involving both complete and partial state measurements. The results obtained indicate the potential of the proposed approach in identifying governing physics from output only measurement.